Bacteria and plants are known to produce cytotoxic proteins which may consist of one, two or several polypeptides or subunits. Those proteins having a single subunit may be loosely classified as Type I proteins. Many of the cytotoxins which have evolved two subunit structures are referred to as type II proteins (Saelinger, C. B. in Trafficking of Bacterial Toxins (eds. Saelinger, C. B.) 1-13 (CRC Press Inc., Boca Raton, Fla., 1990). One subunit, the A chain, possesses the toxic activity whereas the second subunit, the B chain, binds cell surfaces and mediates entry of the toxin into a target cell. A subset of these toxins kill target cells by inhibiting protein biosynthesis. For example, bacterial toxins such as diphtheria toxin or Pseudomonas exotoxin inhibit protein synthesis by inactivating elongation factor 2. Plant toxins such as ricin, abrin, and bacterial toxin Shiga toxin, inhibit protein synthesis by directly inactivating the ribosomes (Olsnes, S. & Phil, A. in Molecular action of toxins and viruses (eds. Cohen, P. & vanHeyningen, S.) 51-105 Elsevier Biomedical Press, Amsterdam, 1982).
Ricin, derived from the seeds of Ricinus communis (castor oil plant), may be the most potent of the plant toxins. It is estimated that a single ricin A chain is able to inactivate ribosomes at a rate of 1500 ribosomes/minute. Consequently, a single molecule of ricin is enough to kill a cell (Olsnes, S. & Phil, A. in Molecular action of toxins and viruses (eds. Cohen, P. & vanHeyningen, S.) (Elsevier Biomedical Press, Amsterdam, 1982). The ricin toxin is a glycosylated heterodimer consisting of A and B chains with molecular masses of 30,625 Da and 31,431 Da linked by a disulphide bond. The A chain of ricin has an N-glycosidase activity and catalyzes the excision of a specific adenine residue from the 28S rRNA of eukaryotic ribosomes (Endo, Y. & Tsurugi, K. J., Biol. Chem. 262:8128 (1987)). The B chain of ricin, although not toxic in itself, promotes the toxicity of the A chain by binding to galactose residues on the surface of eukaryotic cells and stimulating receptor-mediated endocytosis of the toxin molecule (Simmons et al., Biol. Chem. 261:7912 (1986)). Once the toxin molecule consisting of the A and B chains is internalized into the cell via clathrin-dependent or independent mechanisms, the greater reduction potential within the cell induces a release of the active A chain, eliciting its inhibitory effect on protein synthesis and its cytotoxicity (Emmanuel, F. et al., Anal. Biochem. 173: 134-141 (1988); Blum, J. S. et al., J. Biol. Chem. 266: 22091-22095 (1991); Fiani, M. L. et al., Arch. Biochem. Biophys. 307: 225-230 (1993)). Empirical evidence suggests that activated toxin (e.g. ricin, shiga toxin and others) in the endosomes is transcytosed through the trans-Golgi network to the endoplasmic reticulum by retrograde transport before the A chain is translocated into the cytoplasm to elicit its action (Sandvig, K. & van Deurs, B., FEBS Lett. 346: 99-102 (1994).
Protein toxins are initially produced in an inactive, precursor form. Ricin is initially produced as a single polypeptide (preproricin) with a 35 amino acid N-terminal presequence and 12 amino acid linker between the A and B chains. The pre-sequence is removed during translocation of the ricin precursor into the endoplasmic reticulum (Lord, J. M., Eur. J. Biochem. 146:403-409 (1985) and Lord, J. M., Eur. J. Biochem. 146:411-416 (1985)). The proricin is then translocated into specialized organelles called protein bodies where a plant protease cleaves the protein at a linker region between the A and B chains (Lord, J. M. et al., FASAB Journal 8:201-208 (1994)). The two chains, however, remain covalently attached by an interchain disulfide bond (cysteine 259 in the A chain to cysteine 4 in the B chain) and mature disulfide linked ricin is stored in protein bodies inside the plant cells. The A chain is inactive in proricin (O'Hare, M. et al., FEBS Lett. 273:200-204 (1990)) and it is inactive in the disulfide-linked mature ricin (Richardson, P. T. et al., FEBS Lett. 255:15-20 (1989)). The ribosomes of the castor bean plant are themselves susceptible to inactivation by ricin A chain; however, as there is no cell surface galactose to permit B chain recognition the A chain cannot re-enter the cell. The exact mechanism of A chain release and activation in target cell cytoplasm is not known (Lord, J. M. et al., FASAB Journal 8:201-208 (1994)). However, it is known that for activation to take place the disulfide bond between the A and B chains must be reduced and, hence, the linkage between subunits broken.
Diphtheria toxin is produced by Corynebacterium diphtheriae as a 535 amino acid polypeptide with a molecular weight of approximately 58 kD (Greenfield, L. et al., Proc. Natl. Acad. Sci. USA 80:6853-6857 (1983); Pastan, I. et al., Annu. Rev. Biochem. 61:331-354 (1992); Collier, R J. & Kandel, J., J. Biol. Chem. 246:1496-1503 (1971)). It is secreted as a single-chain polypeptide consisting of 2 functional domains. Similar to proricin, the N-terminal domain (A-chain) contains the cytotoxic moiety whereas the C-terminal domain (B-chain) is responsible for binding to the cells and facilitates toxin endocytosis. Conversely, the mechanism of cytotoxicity for diphtheria toxin is based on ADP-ribosylation of EF-2 thereby blocking protein synthesis and producing cell death. The 2 functional domains in diphtheria toxin are linked by an arginine-rich peptide sequence as well as a disulphide bond. Once the diphtheria toxin is internalized into the cell, the arginine-rich peptide linker is cleaved by trypsin-like enzymes and the disulphide bond (Cys 186-201) is reduced. The cytotoxic domain is subsequently translocated into the cytosol substantially as described above for ricin and elicits ribosomal inhibition and cytotoxicity.
Pseudomonas exotoxin is also a 66 kD single-chain toxin protein secreted by Pseudomonas aeruginosa with a similar mechanism of cytotoxicity to that of diphtheria toxin (Pastan, I. et al., Annu. Rev. Biochem. 61:331-354 (1992); Ogata, M. et al., J. Biol. Chem. 267:25396-25401 (1992); Vagil, M. L. et al., Infect. Immunol. 16:353-361 (1977)). Pseudomonas exotoxin consists of 3 conjoint functional domains. The first domain Ia (amino acids 1-252) is responsible for cell binding and toxin endocytosis, a second domain II (amino acids 253-364) is responsible for toxin translocation from the endocytic vesicle to the cytosol, and a third domain III (amino acids 400-613) is responsible for protein synthesis inhibition and cytotoxicity. After Pseudomonas exotoxin enters the cell, the liberation of the cytotoxic domain is effected by both proteolytic cleavage of a polypeptide sequence in the second domain (near Arg 279) and the reduction of the disulphide bond (Cys 265-287) in the endocytic vesicles. In essence, the overall pathway to cytotoxicity is analogous to diphtheria toxin with the exception that the toxin translocation domain in Pseudomonas exotoxin is structurally distinct.
Other toxins possessing distinct functional domains for cytotoxicity and cell binding/toxin translocation include abrin, modeccin and volkensin (Sandvig, K. et al., Biochem. Soc. Trans. 21:707-711 (1993)). Some toxins such as Shiga toxin and cholera toxin also have multiple polypeptide chains responsible for receptor binding and endocytosis.
The ricin gene has been cloned and sequenced, and the X-ray crystal structures of the A and B chains have been described (Rutenber, E. et al. Proteins 10:240-250 (1991); Weston et al., Mol. Bio. 244:410-422, 1994; Lamb and Lord, Eur. J. Biochem. 14:265 (1985); Halling, K. et al. Nucleic Acids Res. 13:8019 (1985)). Similarly, the genes for diptheria toxin and Pseudomonas exotoxin have been cloned and sequenced, and the 3-dimensional structures of the toxin proteins have been elucidated and described (Columblatti, M. et al., J. Biol. Chem. 261:3030-3035 (1986); Allured, V. S. et al., Proc. Natl. Acad. Sci. USA 83:1320-1324 (1986); Gray, G. L. et al., Proc. Natl. Acad. Sci. USA 81:2645-2649 (1984); Greenfield, L. et al., Proc. Natl. Acad. Sci. USA 80:6853-6857 (1983); Collier, R J. et al., J. Biol. Chem. 257:5283-5285 (1982)).
The potential of bacterial and plant toxins for inhibiting mammalian retroviruses, particularly acquired immunodeficiency syndrome (AIDS), has been investigated. Bacterial toxins such as Pseudomonas exotoxin-A and subunit A of diphtheria toxin; dual chain ribosomal inhibitory plant toxins such as ricin, and single chain ribosomal inhibitory proteins such as trichosanthin and pokeweed antiviral protein have been used for the elimination of HIV infected cells (Olson et al., AIDS Res. and Human Retroviruses 7:1025-1030 (1991)). The high toxicity of these toxins for mammalian cells, combined with a lack of specificity of action poses a major problem to the development of pharmaceuticals incorporating the toxins, such as immunotoxins.
Due to their extreme toxicity there has been much interest in making ricin-based immunotoxins as therapeutic agents for specifically destroying or inhibiting infected or tumourous cells or tissues (Vitetta et al., Science 238:1098-1104(1987)). An immunotoxin is a conjugate of a specific cell binding component, such as a monoclonal antibody or growth factor and the toxin in which the two protein components are covalently linked. Generally, the components are chemically coupled. However, the linkage may also be a peptide or disulfide bond. The antibody directs the toxin to cell types presenting a specific antigen thereby providing a specificity of action not possible with the natural toxin. Immunotoxins have been made both with the entire ricin molecule (i.e. both chains) and with the ricin A chain alone (Spooner et al., Mol. Immunol. 31:117-125, (1994)).
Immunotoxins made with the ricin dimer (IT-Rs) are more potent toxins than those made with only the A chain (IT-As). The increased toxicity of IT-Rs is thought to be attributed to the dual role of the B chains in binding to the cell surface and in translocating the A chain to the cytosolic compartment of the target cell (Vitetta et al., Science 238:1098-1104 (1987); Vitetta & Thorpe, Seminars in Cell Biology 2:47-58 (1991)). However, the presence of the B chain in these conjugates also promotes the entry of the immunotoxin into nontarget cells. Even small amounts of B chain may override the specificity of the cell-binding component as the B chain will bind nonspecifically to galactose associated with N-linked carbohydrates, which is present on most cells. IT-As are more specific and safer to use than IT-Rs. However, in the absence of the B chain the A chain has greatly reduced toxicity. Due to the reduced potency of IT-As as compared to IT-Rs, large doses of IT-As must be administered to patients. The large doses frequently cause immune responses and production of neutralizing antibodies in patients (Vitetta et al., Science 238:1098-1104 (1987)). IT-As and IT-Rs both suffer from reduced toxicity as the A chain is not released from the conjugate into the target cell cytoplasm.
A number of immunotoxins have been designed to recognize antigens on the surfaces of tumour cells and cells of the immune system (Pastan et al., Annals New York Academy of Sciences 758:345-353 (1995)). A major problem with the use of such immunotoxins is that the antibody component is its only targeting mechanism and the target antigen is often found on non-target cells (Vitetta et al., Immunology Today 14:252-259 (1993)). Also, the preparation of a suitable specific cell binding component may be problematic. For example, antigens specific for the target cell may not be available and many potential target cells and infective organisms can alter their antigenic make up rapidly to avoid immune recognition. In view of the extreme toxicity of proteins such as ricin, the lack of specificity of the immunotoxins may severely limit their usefulness as therapeutics for the treatment of cancer and infectious diseases.
The insertion of intramolecular protease cleavage sites between the cytotoxic and cell-binding components of a toxin can mimic the way that the natural toxin is activated. European patent application no. 466,222 describes the use of maize-derived pro-proteins which can be converted into active form by cleavage with extracellular blood enzymes such as factor Xa, thrombin or collagenase. Garred, O. et al. (J. Biol. Chem. 270:10817-10821 (1995)) documented the use of a ubiquitous calcium-dependent serine protease, furin, to activate shiga toxin by cleavage of the trypsin-sensitive linkage between the cytotoxic A-chain and the pentamer of cell-binding B-units. Westby et al. (Bioconjugate Chem. 3:375-381 (1992)) documented fusion proteins which have a specific cell binding component and proricin with a protease sensitive cleavage site specific for factor Xa within the linker sequence. O'Hare et al. (FEBS Lett. 273:200-204 (1990)) also described a recombinant fusion protein of RTA and staphylococcal protein A joined by a trypsin-sensitive cleavage site. In view of the ubiquitous nature of the extracellular proteases utilized in these approaches, such artificial activation of the toxin precursor or immunotoxin does not confer a mechanism for intracellular toxin activation and the problems of target specificity and adverse immunological reactions to the cell-binding component of the immunotoxin remain.
In a variation of the approach of insertion of intramolecular protease cleavage sites on proteins which combine a binding chain and a toxic chain, Leppla, S. H. et al. (Bacterial Protein Toxins zbl.bakt.suppl. 24:431-442 (1994)) suggest the replacement of the native cleavage site of the protective antigen (PA) produced by Bacillus anthracis with a cleavage site that is recognized by cells that contain a particular protease. PA, recognizes, binds, and thereby assists in the internalization of lethal factor (LF) and edema toxin (ET). also produced by Bacillus anthracis. However, this approach is wholly dependent on the availability of LF, or ET and PA all being localized to cells wherein the modified PA can be activated by the specific protease. It does not confer a mechanism for intracellular toxin activation and presents a problem of ensuring sufficient quantities of toxin for internalization in target cells.
The in vitro activation of a Staphylococcus-derived pore-forming toxin, α-hemolysin by extracellular tumour-associated proteases has been documented (Panchel, R. G. et al., Nature Biotechnology 14:852-857 (1996)). Artificial activation of α-hemolysin in vitro by said proteases was reported but the actual activity and utility of α-hemolysin in the destruction of target cells were not demonstrated.
Hemolysin does not inhibit protein synthesis but is a heptameric transmembrane pore which acts as a channel to allow leakage of molecules up to 3 kD thereby disrupting the ionic balances of the living cell. The α-hemolysin activation domain is likely located on the outside of the target cell (for activation by extracellular proteases). The triggering mechanism in the disclosed hemolysin precursor does not involve the intracellular proteolytic cleavage of 2 functionally distinct domains. Also, the proteases used for the α-hemolysin activation are ubiquitously secreted extracellular proteases and toxin activation would not be confined to activation in the vicinity of diseased cells. Such widespread activation of the toxin does not confer target specificity and limits the usefulness of said α-hemolysin toxin as therapeutics due to systemic toxicity.
A variety of proteases specifically associated with malignancy, viral infections and parasitic infections have been identified and described. For example, cathepsin is a family of serine, cysteine or aspartic endopeptidases and exopeptidases which has been implicated to play a primary role in cancer metastasis (Schwartz, M. K., Clin. Chim. Acta 237:67-78 (1995); Spiess, E. et al., J. Histochem. Cytochem. 42:917-929 (1994); Scarborough, P. E. et al., Protein Sci. 2:264-276 (1993); Sloane, B. F. et al., Proc. Natl. Acad. Sci. USA 83:2483-2487 (1986); Mikkelsen, T. et al., J. Neurosurge 83:285-290 (1995)). Matrix metalloproteinases (MMPs or matrixins) are zinc-dependent proteinases consisting of collagenases, matrilysin, stromelysins, gelatinases and macrophage elastase (Krane, S. M., Ann. N.Y. Acad. Sci. 732:1-10 (1994); Woessner, J. F., Ann. N.Y. Acad. Sci. 732:11-21 (1994); Carvalho, K. et al., Biochem. Biophys. Res. Comm. 191:172-179 (1993); Nakano, A. et al. J. of Neurosurge, 83:298-307 (1995); Peng, K-W, et al. Human Gene Therapy, 8:729-738 (1997); More, D. H. et al. Gynaecologic Oncology, 65:78-82 (1997)). These proteases are involved in pathological matrix remodeling. Under normal physiological conditions, regulation of matrixin activity is effected at the level of gene expression. Enzymatic activity is also controlled stringently by tissue inhibitors of metalloproteinases (TIMPs) (Murphy, G. et al., Ann. N.Y. Acad. Sci. 732:31-41 (1994)). The expression of MMP genes is reported to be activated in inflammatory disorders (e.g. rheumatoid arthritis) and malignancy.
In malaria, parasitic serine and aspartic proteases are involved in host erythrocyte invasion by the Plasmodium parasite and in hemoglobin catabolism by intraerythrocytic malaria (O'Dea, K. P. et al., Mol. Biochem. Parasitol. 72:111-119 (1995); Blackman, M. J. et al., Mol. Biochem. Parasitol. 62:103-114 (1993); Cooper, J. A. et al., Mol. Biochem. Parasitol. 56:151-160 (1992); Goldberg, D. E. et al., J. Exp. Med. 173:961-969 (1991)). Schistosoma mansoni is also a pathogenic parasite which causes schistosomiasis or bilharzia. Elastinolytic proteinases have been associated specifically with the virulence of this particular parasite (McKerrow, J. H. et al., J. Biol. Chem. 260:3703-3707 (1985)).
Welch, A. R. et al. (Proc. Natl. Acad. Sci. USA 88:10797-10800 (1991)) has described a series of viral proteases which are specifically associated with human cytomegalovirus, human herpesviruses, Epstein-Barr virus, varicella zoster virus-I. and infectious laryngotracheitis virus. These proteases possess similar substrate specificity and play an integral role in viral scaffold protein restructuring in capsid assembly and virus maturation. Other viral proteases serving similar functions have also been documented for human T-cell leukemia virus (Blaha, I. et al., FEBS Lett. 309:389-393 (1992); Pettit, S. C. et al., J. Biol. Chem. 266:14539-14547 (1991)), hepatitis viruses (Hirowatari, Y. et al., Anal. Biochem. 225:113-120 (1995); Hirowatari, Y. et al., Arch. Virol. 133:349-356 (1993); Jewell, D. A. et al., Biochemistry 31:7862-7869 (1992)), poliomyelitis virus (Weidner, J. R. et al., Arch. Biochem. Biophys. 286:402-408 (1991)), and human rhinovirus (Long, A. C. et al., FEBS Lett. 258:75-78 (1989)).
Candida yeasts are dimorphic fungi which are responsible for a majority of opportunistic infections in AIDS patients (Holmberg, K. and Myer, R., Scand. J. Infect. Dis. 18:179-192 (1986)). Aspartic proteinases have been associated specifically with numerous virulent strains of Candida including Candida albican, Candida tropicalis, and Candida parapsilosis (Abad-Zapatero, C. et al., Protein Sci. 5:640-652 (1996); Cutfield, S. M. et al., Biochemistry 35:398-410 (1995); Ruchel, R. et al, Zentralbl. Bakteriol. Mikrobiol Hyg. I Abt. Orig. A. 255:537-548 (1983); Remold, H. et al., Biochim. Biophys. Acta 167:399-406 (1968)), and the levels of these enzymes have been correlated with the lethality of the strain (Schreiber, B, et al., Diagn. Microbiol. Infect. Dis. 3:1-5 (1985)).